finding(devx): vitest.config.mts is walked by NO ESLint run in this repository -- .mts/.cts are in neither the rule-bearing globs nor ESLint's default lint set

Split out of objectui#7908, which measured the files ESLint walks with an empty rule set. This is the neighbouring class it cannot see: a file ESLint never walks at all.

The reading

Measured on fedfa3e4a with a real install, ESLint v10.8.1, via the ESLint Node API (isPathIgnored / calculateConfigForFile, the API half of --print-config):

vitest.config.mts   ->  isPathIgnored: true, config: undefined   (NOT WALKED)
playwright.config.ts ->  116 rules                               (control)
scripts/github-slug.mjs ->  0 rules, but WALKED                  (control, the #7908 class)

Three controls in one run, and the three answers are three different states. undefined is not "zero rules" — it is ESLint declining to look.

Why

In flat config a file is linted only if some config object's files matches it, plus the default set ESLint always lints, which is .js / .cjs / .mjs. Every rule-bearing object in eslint.config.js is scoped TS-and-TSX — spelled as a brace-expansion over those two extensions — and .mts / .cts are in neither that glob nor the default set. So the extension falls through both.

The population

One file today, repo-wide: vitest.config.mts. There are no .cts files. Counted with the same walk objectui#7908's gate uses: 4440 files walked, and this one is outside all of them.

It is not a big population, but it is not a trivial file either — it defines every project, include glob and setup file the entire test run uses, and it is the one config in this repository that no lint run has ever read.

Why this is filed rather than repaired

objectui#7908's dispatch scoped that card to making vacuity visible and explicitly left rule-STRENGTH decisions out. Widening the config's files to reach .mts is a strength decision of exactly that kind: it is one line, but its red set is UNMEASURED, and the same objection applies as for the .js family. It also cannot be caught by #7908's gate by construction — that gate reports files ESLint walks, and this file is not one of them.

Two ways to settle it

Widen the glob — add .mts / .cts to the rule-bearing files entries, measure the red set first.
Gate the gap — extend check:lint-rule-coverage with a second predicate over source extensions ESLint reaches for NEITHER reason, so a .mts or .cts added later is reported instead of being silently outside every run. That is a real widening of that gate's subject and deserves its own decision.

Filed unassigned by the objectui#7908 implementation session. Generated by Claude Code, session session_01FhBNJcLRZLe8M87VcUgpKr.

Generated by Claude Code

Triage: lands in domain:devx (eslint.config.js / check:lint-rule-coverage); rationale: class (a) — ⭐ three controls in one run yielding three distinct states, which is the cleanest discrimination in today's batch:

vitest.config.mts        → isPathIgnored: true, config: undefined   NOT WALKED
playwright.config.ts     → 116 rules                                 walked + ruled
scripts/github-slug.mjs  → 0 rules, but WALKED                       the #7908 class

⭐ "undefined is not 'zero rules' — it is ESLint declining to look." ⇒ That sentence is the whole finding, and it is what makes this a different defect from objectui#7908 rather than more of it. Bug.

The cause is read off the flat-config rules rather than inferred: a file is linted only if some config object's files matches, plus ESLint's always-linted default set (.js / .cjs / .mjs). Every rule-bearing object here is scoped to TS-and-TSX, and .mts / .cts are in neither the glob nor the default set. ⇒ The extension falls through both.

priority:p3, and the population is why: one file, repo-wide (no .cts files exist), counted with the same 4440-file walk #7908's gate uses. ⚠️ But it is not a trivial file — "it defines every project, include glob and setup file the entire test run uses, and it is the one config in this repository that no lint run has ever read."

⚠️ ⭐ And it cannot be caught by objectui#7908's gate, by construction — that gate reports files ESLint walks, and this file is not one of them. ⇒ Landing #7908 (graded p2 today) does not close this, and ⛔ nobody should read that card's completion as covering .mts.

Route: prefer option 2 — gate the gap. Extend check:lint-rule-coverage with a second predicate over source extensions ESLint reaches for neither reason, so a .mts or .cts added later is reported rather than silently outside every run. ⇒ That scales with the repo and needs no red-set measurement to be safe. ⚠️ It is a real widening of that gate's subject and deserves saying so explicitly in the PR.

⚠️ Option 1 (widen the glob) requires measuring the red set first, and ⛔ must not be taken as "one line, obviously right": the same objection #7908's dispatch raised for the .js family applies — the red set is unmeasured, and turning rules on for a config file nothing has ever linted is a rule-strength decision, not a coverage fix. ⇒ If both are done, the gate lands first so the widening's red set is visible before it is chosen.

⛔ Correctly split from objectui#7908 rather than folded in: that card's dispatch scoped it to making vacuity visible and explicitly left rule-strength decisions out.

⛔ This seat grades and routes only: not claimed, not dispatched, no code.

Generated by Claude Code

Claim: PM loop round R46 — domain:devx @ objectui execution seat.
Session: session_01FhBNJcLRZLe8M87VcUgpKr
Branch: claude/issue-8337-mts-lint-coverage
Worktree: /home/user/objectui-issue-8337
Domain: domain:devx
Dispatch: one os-dev subagent of this session (mode:subagent, model: opus); the dispatch note with the brief follows. Labels pm:queue → pm:dispatched and assignee set in one write, read back and diffed against the union at 2026-09-08T01:12Z.

⚠️ Five cards this shift carried a false premise inside the card face itself — #8006 (line numbers drifted twice), #8029 (the card and its triage comment cited a file that does not exist), #8113 (the card claimed the rest of the page was correct; three things on it were wrong), #8117 (card and triage both prescribed deJsxClosingTags, retired by #7883 and not callable), #8122 (its line coordinate was 52 lines off). ⇒ This dispatch requires the defect to be re-derived from the tree before any edit, and treats a corrected card with zero implementation as a good outcome.

⚠️ And five probe expectations would have mis-scored a correct answer today — the most recent two were this seat's own: a stale census figure carried out of its own seat post into a dispatch brief, and a gate run in the shared checkout instead of on origin/main. ⛔ The instrument is the first suspect, and a zero-hit reading means nothing without a control.

Generated by Claude Code

Dispatched — domain:devx @ objectui execution seat, PM session session_01FhBNJcLRZLe8M87VcUgpKr, R46, 2026-09-08T01:16Z. One of three slots (with #7925 and #8047).

Dev. One os-dev subagent of this session on branch claude/issue-8337-mts-lint-coverage, worktree /home/user/objectui-issue-8337, branching from origin/main at 868e82501 or later. Race re-read at 01:13Z: the newest Claim: is 5577437750, naming that branch — and it is the only one. ⛔ The dev posts no second claim and ⛔ never writes the assignee.

Premise re-verified by this seat before dispatch. On origin/main, every rule-bearing object in eslint.config.js is scoped to TS-and-TSX, and .mts / .cts are in neither that glob nor ESLint's always-linted default set (.js / .cjs / .mjs). ⇒ vitest.config.mts falls through both. The card's own three-control run is what makes this readable, and the brief requires it re-taken rather than quoted:

vitest.config.mts        -> isPathIgnored: true, config: undefined   NOT WALKED
playwright.config.ts     -> 116 rules                                walked + ruled
scripts/github-slug.mjs  -> 0 rules, but WALKED                      the #7908 class

⭐ Three controls in one run yielding three distinct states — and triage is right that the sentence "undefined is not 'zero rules' — it is ESLint declining to look" is the whole finding. ⚠️ It is also why landing objectui#7908 does not close this: that gate reports files ESLint walks, and this file is not one of them. ⛔ Nobody should read #7908's completion as covering .mts.

Triage's route taken as given, with its warning attached. Option 2 — extend check:lint-rule-coverage with a second predicate over source extensions ESLint reaches for neither reason — so a .mts or .cts added later is reported instead of silently outside every run. ⚠️ Triage says out loud that this is "a real widening of that gate's subject and deserves saying so explicitly in the PR", and the brief carries that. ⛔ Option 1 (widen the glob) is not a one-liner: it requires the red set measured first, and it stays out of this PR even if the red set turns out empty. If the dev's measurement says option 1 is the better answer, ⛔ it does not switch silently — it reports the conflict and this seat routes it.

⛔ Two premises re-read rather than inherited. 4440 is the card's file-walk count, not today's. And objectui#7908's own state is a premise: whether it has landed decides what check:lint-rule-coverage looks like in the worktree, so the new predicate sits beside what is actually there.

⚠️ The failure this dispatch is most worried about is a gate that is green because its population is empty. The brief makes non-vacuity non-optional in both directions — red on an injected .mts/.cts source, green on the files that are walked — with the mutation proven on disk by hash and marker count before any result is read, restored by state under an EXIT INT TERM trap.

⚠️ And the four-axis block is pointed at the deliverable itself. A new gate predicate is a capability expansion and the fourth axis defaults against it. Triage has argued the pull; the dev must still weigh it, and "the predicate is not worth its maintenance" is an admissible answer here if the measurement says so.

Generated by Claude Code

os-dev-report

{
  "issue": 8337,
  "status": "done",
  "branch": "claude/issue-8337-mts-lint-coverage",
  "pr": "https://github.com/objectstack-ai/objectui/pull/8467",
  "pr_number": 8467,
  "head_sha": "8e75df1cdf3d1bd7ac6cb8c32567ee9457b5c775",
  "base": "868e825012b7c93e4df36c4bf9ab8d8dc55dcb02",
  "premise_still_valid": true,
  "premise_notes": "Re-derived on 868e825012 with a real install, ESLint v10.8.1, via the Node API before any edit. The three controls reproduce as three DISTINCT states, not collapsed: vitest.config.mts -> isPathIgnored true / calculateConfigForFile undefined (NOT WALKED); playwright.config.ts -> not ignored, 116 rules; scripts/github-slug.mjs -> not ignored, 0 rules. Corroborated from lint:root's own --format json output: 275 files linted, vitest.config.mts in none of them. Population re-taken, NOT carried from the card: 4480 files walked today (the card's 4440 is stale), one .mts file repo-wide, zero .cts. Two card-face corrections found: (1) the unreachable extension set is THREE, not one -- .jsx is unreachable for exactly the same reason and neither the card nor its triage names it (zero .jsx files today, so nothing reported it and nothing would have); (2) objectui#7908 has LANDED (closed 2026-09-07 by PR #8339), so check:lint-rule-coverage already existed and predicate 2 was written to sit beside what is actually in the tree.",
  "summary": "Took triage's option 2 -- gated the gap -- by adding a second predicate to check:lint-rule-coverage: every source file ESLint does NOT walk must be unwalked for some reason other than its extension. The unreachable extension set is DERIVED from the live config on every run (one synthetic probe per source extension) rather than listed, so it is not a .mts special case; per file the discrimination is a substitution on the candidate's own path, which separates the extension gap from location and name exclusions. One walk feeds both predicates (isPathIgnored is 1.9s of the gate's ~2s, so walking twice would double the cost). vitest.config.mts is a declared row in a second shrink-only ledger whose rows are exact PATHS, not population globs -- a **/*.mts row would waive the only thing the predicate exists to catch, and that asymmetry with VACUOUS_GROUPS is pinned by a test. The gate's own header now states that this PR widened its subject and why predicate 1 cannot see predicate 2's population by construction. Option 1 was NOT taken and NOT switched to silently; its red set was measured and handed on as a separate card. The PR is draft and stays draft.",
  "files_changed": [
    "scripts/check-lint-rule-coverage.mjs",
    "scripts/__tests__/check-lint-rule-coverage.test.ts"
  ],
  "tests": "Suite: `pnpm exec vitest run scripts/__tests__/check-lint-rule-coverage.test.ts` under the shared lock -- `Test Files 1 passed (1)`, `Tests 27 passed (27)` (20 before, 7 added); wrapper `VERDICT command-exit 0 held the lock 5s waited 0s`. Lane: `pnpm lint:root` run VERBATIM from package.json, nothing narrowed -- wrapper `VERDICT command-exit 0`, `32 problems (0 errors, 32 warnings)`, all warnings pre-existing and none in the two changed files; a second run with `--format json` gives the population as 275 files linted, both changed files present, vitest.config.mts absent. Gates, exit codes captured by redirect-then-capture, never through a pipe: check-changeset-presence EXIT 0 (`no changeset is owed` -- run, not guessed); check-governed-queue-guard --test on both paths EXIT 0 (`NOT GOVERNED -- 2 path(s) checked against 5 governed surface(s)`); check-control-bytes EXIT 0 plus a direct `grep -naP` control-character scan over both changed files with zero hits; check-entry-guard EXIT 0. ABLATION -- five legs, all from the COMMITTED tree, each proving the mutation reached disk by hash AND marker count BEFORE any result was read, each under `trap ... EXIT INT TERM` with absolute paths, each restore proven BY STATE (`git diff HEAD` empty) and never by an exit code. (A) gate BEFORE: base blob b2b3e4bd checked out and hash-matched on disk, EXIT=0, zero occurrences of vitest.config.mts in its output -- the defect is invisible to the gate as it stands. (B) non-vacuity: inject scripts/ablation-injected-tool.cts -> EXIT=1 naming it. (C) negative control: identical bytes, identical directory, .ts instead -> EXIT=0, walked count 4480 goes to 4481, so B's red is about the extension and not the file. (D) location control: identical bytes as .cts inside packages/core/dist/ -> EXIT=0, not reported. (E) remedy direction: widen the two rule-bearing globs to reach .mts -> EXIT=1 in BOTH shrink-only directions (unreached-over-broad naming vitest.config.mts, and unreached-stale printing `Delete the row`), so the ledger row cannot become a permanent exemption. Final tree state after all legs: `git status --porcelain` empty, `git diff HEAD` empty. NOT RUN LOCALLY, declared to CI: the full `pnpm test` suite and the repository-wide check:* farm -- this diff touches two files under scripts/ and the affected suite is the gate's own pin test, which was run.",
  "four_axis_analysis": "The full block is in the PR body in Chinese, as the axis set requires. Verdict: BUILD it, with the two costs stated rather than hidden. (1) 实际业务需求 -- measured, not asserted: the population today is one real and load-bearing file (vitest.config.mts defines every project, include glob and setup the whole test run uses) and it is absent from all 275 files lint:root walks. Honest half: .cts and .jsx have zero files, so a substantial part of this predicate's value is prospective. (2) 项目长远合理性 -- contract-first: the alternative is a rule-strength change with (then) an unmeasured red set; this route makes the gap VISIBLE without deciding strength, the ledger is shrink-only, and leg E proves the remedy mechanically reds the waiver and names the deletion. No consumer-side leniency, no ?? fallback. (3) 防 AI 写代码犯错 -- the strongest axis here, and it decides the case: an AI adding vitest.config.cts or a .mts build script today gets feedback in NEITHER direction; eslint . never opens the file and exits 0, and every downstream reading calls that clean. That is precisely the declared-but-not-enforced shape this axis bans, and the change converts silence into a loud, file-naming rejection with the two legitimate exits spelled out in the failure text. (4) 创业阶段不扩散需求 -- this axis defaults AGAINST a new gate and the objection partly stands (population of 1). Against it: this is not a new gate but ~90 lines inside an existing one -- no new script, no new workflow, no new CI minutes, reusing the same walk (marginal cost is 9 extension probes plus one substitution probe per candidate); it declares no new authorable capability and no new metadata surface; the maintenance face is one ledger row that can only shrink; and it grants no transition window (exact path today, not a **/*.mts grace period). No genuine axis conflict; the tension is (1)/(4) against (3), and (3) wins because the silent-mass-error shape it guards is one this repository has already measured happening.",
  "mcp_calls": "8 -- issue_read x3 (8337 get, 7908 get, 8337 comments), create_pull_request x1, pull_request_read x1 (body read-back), issue_write x2 (the two findings), add_issue_comment x1 (this report). Dedup for both filed cards went through the ZERO-quota channel instead: repo-scoped REST probed first (HTTP 200), then two label-scoped REST listings (tooling, finding) plus a local grep, 118 distinct open issues scanned, with #8337 itself as the known-present control in the same run -- CONTROL HIT, so the zero-duplicate reading is a reading.",
  "open_questions": [
    {
      "question": "The harness attribution template injected into this session ends commit messages with a Co-Authored-By trailer whose display name is a model identifier; the os-dev dispatch contract carries, verbatim, `never put a model identifier in any pushed artifact`. A commit message is a pushed artifact, so the two cannot both be satisfied. Which gives way?",
      "options": [
        "A -- keep the trailer shape and the session line, write the co-author display name as the bare product name with no model designation (what this seat did, acting on the stricter instruction)",
        "B -- follow the harness template verbatim and treat the dispatch ban as scoped to product-visible artifacts (PR bodies, issue text, shipped docs) rather than to commit trailers",
        "C -- drop the Co-Authored-By trailer entirely and keep only the Claude-Session line"
      ],
      "recommendation": "A for now, and it is what commit 8e75df1cd does -- it is the only option that satisfies the strict reading of the ban while preserving attribution. But A is a per-seat judgement call that fires on EVERY commit of EVERY dispatch, so seats will resolve it differently and main ends up with three shapes in one machine-readable field. The durable fix is B or C spelled out once in the dispatch contract, and that is a maintainer's call on governed-surface text, not this seat's. Filed as #8471."
    },
    {
      "question": "Widening the rule-bearing globs (triage's option 1) now has a measured red set, and it is EMPTY -- with mts,cts added, vitest.config.mts resolves 116 rules and `eslint vitest.config.mts` exits 0 with no output. Does that change who decides it, or only what they decide with?",
      "options": [
        "A -- only what they decide with: the objection that blocked option 1 was the missing number, but turning rules on for a never-linted population is still a strength decision with its own owner",
        "B -- the empty red set makes it a coverage fix rather than a strength change, so it can ride an ordinary tooling PR"
      ],
      "recommendation": "A, and this PR acts on A: option 1 was measured but NOT taken, and NOT switched to silently. Filed as #8469 so whoever owns strength decisions has the number instead of the objection. Note for that decision: the population is three extensions, not two -- .jsx is unreachable for the same reason and any widening should decide about all three."
    }
  ],
  "out_of_scope_findings": [
    "filed as #8469: option 1 on objectui#8337 (widen the rule-bearing globs to .mts/.cts) has an EMPTY red set -- measured under trap-and-restore on 868e825012, 116 rules resolved and eslint exits 0 with no output; NOT taken here because it remains a rule-strength decision, and the population is three extensions once .jsx is counted",
    "filed as #8471: the harness attribution template and the os-dev ban on model identifiers in pushed artifacts contradict each other in every commit message; acted on the stricter instruction and filed rather than picking a side silently"
  ]
}

Draft PR #8467 is open and stays draft. The governed-surface guard reports NOT GOVERNED for both changed paths, so the draft here is the dispatch's instruction rather than the guard's; CI convergence is this seat's to watch, not mine.

Generated by Claude Code; the session reference is session_01FhBNJcLRZLe8M87VcUgpKr, written as prose deliberately.

Generated by Claude Code

ACCEPT — domain:devx @ objectui execution seat, PM session session_01FhBNJcLRZLe8M87VcUgpKr, R46, reviewed 2026-09-08T01:42Z against GitHub and the tree, ⛔ never against the report's narrative.

⭐ The dev found a third unreachable extension that neither the card nor triage names

The card scopes the defect to .mts / .cts. Re-derived here on a clean origin/main extract, every rule-bearing files entry in eslint.config.js:

**/*.{ts,tsx}      **/*.test.{ts,tsx}      **/__tests__/**/*.{ts,tsx}      **/*.tsx      packages/types/src/objectql.ts

⇒ .jsx is named by none of them, and ESLint's always-linted default set is .js / .cjs / .mjs. .jsx is unreachable for exactly the same reason, and nothing has ever reported it because the population is zero today (.mts 1, .cts 0, .jsx 0; control .tsx 2504). ⭐ That is the finding restated one size larger: a listed special case would have shipped the blind spot at a smaller radius, which is why the dev derived the unreachable set from the live config on every run — one synthetic probe per source extension — instead of writing .mts into the gate.

⛔ Two premises re-derived, and both had moved

4440 is stale — the walk returns 4480 today. The dispatch required the number taken, not carried, and it was.
objectui#7908 has LANDED. Verified on GitHub: closed 2026-09-07T14:04:29Z, and scripts/check-lint-rule-coverage.mjs is present on origin/main. ⇒ The dev wrote predicate 2 beside what is actually in the tree rather than beside the card's description of it, which is what the brief asked for and is why this is an extension of an existing gate rather than a new one.

The card's own three controls reproduce as three distinct states, not collapsed — vitest.config.mts ignored with config: undefined, playwright.config.ts at 116 rules, scripts/github-slug.mjs walked with 0 rules — corroborated from lint:root's own --format json population of 275 files, with vitest.config.mts in none of them.

⭐ The ablation is five legs and each one answers a different objection

leg	what it rules out
A — the gate at its base blob, hash-matched on disk	exit 0 with zero occurrences of `vitest.config.mts`: the defect really is invisible to the gate as it stands
B — inject `scripts/ablation-injected-tool.cts`	exit 1, naming it
C — identical bytes, identical directory, `.ts` instead	exit 0, walked count 4480 → 4481 ⇒ B's red is about the extension, not the file
D — identical bytes as `.cts` inside `packages/core/dist/`	exit 0, not reported ⇒ the extension gap is separated from location exclusions
E — widen the two rule-bearing globs to reach `.mts`	exit 1 in both shrink-only directions — `unreached-over-broad` naming the file, and `unreached-stale` printing `Delete the row`

⭐ Leg E is the one that matters most, and it is the answer to the objection this seat would otherwise have raised: a ledger row is a waiver, and a waiver that survives its own remedy is a permanent exemption. It cannot survive — the moment the real fix lands, the gate demands the row's deletion. Each leg proved its mutation reached disk by hash and marker count before any result was read, ran under trap … EXIT INT TERM with absolute paths, and restored by state (git diff HEAD empty), never by exit code.

⭐ And the ledger's rows are exact paths, not population globs, with the asymmetry against VACUOUS_GROUPS pinned by a test — because a **/*.mts row would waive the only thing the predicate exists to catch. That is the difference between a debt list and a hole.

⭐ Option 1 was measured and NOT taken, which is the behaviour the dispatch demanded

The dev measured triage's option 1 under trap-and-restore: with mts,cts added, vitest.config.mts resolves 116 rules and eslint vitest.config.mts exits 0 with no output — the red set is empty. An empty red set is the most tempting possible moment to switch routes silently. It did not: option 1 stayed out of this PR and went to #8469 with the number attached, on the reading that turning rules on for a never-linted population is still a strength decision with its own owner. ⇒ ⛔ Confirmed by this seat: that reading is correct, and the note that any widening must decide about three extensions rather than two belongs with it.

The costs are stated rather than hidden

The four-axis answer concedes the fourth axis partly stands — the live population is one file, and .cts/.jsx value is prospective — and argues it is ~90 lines inside an existing gate, no new script, no new workflow, no new CI minutes, one walk feeding both predicates (isPathIgnored is 1.9s of the gate's ~2s, so walking twice would double it). ⛔ No transition window and no **/*.mts grace period, per 创业阶段不渐进. The gate's own header now records that this PR widened its subject and why predicate 1 cannot see predicate 2's population by construction. ⇒ That is a subject moved on purpose, which is what triage asked to be said out loud.

+590/-33 across the gate and its pin (27 tests, 20 before). lint:root run verbatim from package.json, 275 files, 0 errors, 32 pre-existing warnings, none in the changed files. check-changeset-presence exit 0, none owed, run rather than guessed. check-governed-queue-guard --test: NOT GOVERNED. Both filed findings (#8469, #8471) carry the finding label only, unassigned — ⛔ no half-state to repair.

⚠️ #8471 is a real conflict and the dev resolved it correctly: this session's attribution instruction ends commit messages with a trailer naming a model, while the dispatch contract bans a model identifier in any pushed artifact. It acted on the stricter instruction and filed rather than choosing silently — which is the rule working. ⇒ ⛔ This seat does not answer it either; the durable fix is dispatch-contract text, and that is governed surface.

Next: ⚠️ the queue is held for the moment — see the note this seat is posting about Test (shard 2/4) and the base branch. Once that is settled: flip ready → post-flip guard → auto-merge SQUASH → queue confirmation → landing probe (the second predicate present in scripts/check-lint-rule-coverage.mjs and the unreachable set derived from the config rather than listed, on re-fetched origin/main; control: the vitest.config.mts ledger row present as an exact path and the VACUOUS_GROUPS asymmetry pin still in the test file — they are what stop the row becoming a permanent waiver) → LANDED here and on #8337, which closes via Fixes.

Generated by Claude Code

LANDED — domain:devx @ objectui execution seat, PM session session_01FhBNJcLRZLe8M87VcUgpKr, R46. PR #8467 merged 2026-09-08T01:47:13Z as ed17be480; probe run on a clean extract of re-fetched origin/main, ⛔ never on the shared working tree.

leg	reading on `origin/main@ed17be480`
probe — the unreachable set is DERIVED, not listed	`:196 So the unreachable set is DERIVED from the live config on every run`
probe — ⭐ and the header records the widened finding	`:201 unreachable jsx cts mts` · `:203 ⭐ THREE, not one. \`.jsx` is unreachable here for exactly the same reason`
probe — ⛔ no `.mts` special case in the gate	0 quoted-literal occurrences. ⚠️ Controlled: the string appears in prose at `:134 :162 :181 :227`, so the grep fires — the zero is about the literal, not about the term being absent
control — the ledger row is an exact PATH	`:417 export const UNREACHED_GROUPS = [ { glob: 'vitest.config.mts',` — ⛔ not `*/.mts`, which would waive the only thing the predicate exists to catch
control — the `VACUOUS_GROUPS` asymmetry pin survives	`check-lint-rule-coverage.test.ts:476 / :484 / :492`, including its comment "The asymmetry with VACUOUS_GROUPS, pinned because it is deliberate"

⚠️ The gate itself cannot execute in a bare git archive extract (ERR_MODULE_NOT_FOUND, no node_modules); that is the extract's limit and ⛔ not a reading about the gate. Its behaviour was measured in the dev's worktree and by CI, both green.

⭐ What landed that the card did not ask for: .jsx is unreachable by exactly the same two gaps as .mts/.cts, and neither the card nor its triage comment named it. Zero .jsx files exist today, so nothing had ever reported it — which is why the predicate derives its set from the live config instead of naming extensions. The finding is now in the gate's own header where the next reader hits it.

⛔ Option 1 was measured and not taken: with mts,cts added, vitest.config.mts resolves 116 rules and eslint exits 0 with no output — an empty red set, and the most tempting possible moment to switch routes silently. It went to #8469 with the number attached instead, on the reading that turning rules on for a never-linted population is still a strength decision with its own owner. ⇒ This seat confirms that reading.

pm:dispatched stripped in the same pass; domain:devx, bug, tooling, priority:p3 stay. The card closed via Fixes.

Generated by Claude Code

Navigation Menu

Repository navigation

finding(devx): `vitest.config.mts` is walked by NO ESLint run in this repository -- `.mts`/`.cts` are in neither the rule-bearing globs nor ESLint's default lint set #8337

The reading

Why

The population

Why this is filed rather than repaired

Two ways to settle it

⭐ The dev found a third unreachable extension that neither the card nor triage names

⛔ Two premises re-derived, and both had moved

⭐ The ablation is five legs and each one answers a different objection

⭐ Option 1 was measured and NOT taken, which is the behaviour the dispatch demanded

The costs are stated rather than hidden

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

finding(devx): vitest.config.mts is walked by NO ESLint run in this repository -- .mts/.cts are in neither the rule-bearing globs nor ESLint's default lint set #8337

Description

The reading

Why

The population

Why this is filed rather than repaired

Two ways to settle it

Activity

os-zhuang commented on Sep 7, 2026

claude commented on Sep 8, 2026

claude commented on Sep 8, 2026

baozhoutao commented on Sep 8, 2026

claude commented on Sep 8, 2026

⭐ The dev found a third unreachable extension that neither the card nor triage names

⛔ Two premises re-derived, and both had moved

⭐ The ablation is five legs and each one answers a different objection

⭐ Option 1 was measured and NOT taken, which is the behaviour the dispatch demanded

The costs are stated rather than hidden

claude commented on Sep 8, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

finding(devx): `vitest.config.mts` is walked by NO ESLint run in this repository -- `.mts`/`.cts` are in neither the rule-bearing globs nor ESLint's default lint set #8337